OpenAI Chief Strategy Officer Jason Kwon speaks during a press conference at JW Marriott Hotel Seoul, Wednesday. Courtesy of OpenAI Korea
OpenAI will include the Korean government, public institutions and companies in its cybersecurity program, allowing access to its latest cybersecurity artificial intelligence (AI) models, its chief strategy officer (CSO) said Wednesday.
“We believe AI can become a powerful tool for cyber defenders, helping them identify vulnerabilities faster, respond to threats more effectively and make software and systems more secure,” OpenAI CSO Jason Kwon said during a press conference in Seoul. “We want OpenAI to become a trusted partner to Korea.”
In particular, Korea will participate in OpenAI’s Daybreak cybersecurity initiative. Under the initiative, Korean institutions will gain Governmental Trusted Access for Cyber (GTAC), a restricted-access tier that allows verified cybersecurity professionals to utilize specialized AI models to test cybersecurity vulnerability. Kwon said OpenAI will expand the TAC program to private companies in Korea that are involved in the country’s key industries.
According to OpenAI, Korea’s participation in GTAC marks the first case in Asia alongside Japan, with participating institutions gaining access to the company’s latest high-performance AI model — GPT-5.5-Cyber. In Korea, the Korea Internet & Security Agency will oversee the program in practice.
“Our goal is not to develop and to keep advanced cyber capabilities in the hands of a small number of organizations,” Kwon said. “Our goal is to make them available to more trusted defenders who are responsible for protecting the systems, companies, institutions and people that society depends on.”
Second Vice Minister of Science and ICT Ryu Je-myung, right, poses with OpenAI Chief Strategy Officer Jason Kwon during their meeting in Seoul, Tuesday. Courtesy of Ministry of Science and ICT
Korea’s inclusion in OpenAI’s cybersecurity program was confirmed during a meeting between Kwon and Second Vice Minister of Science and ICT Ryu Je-myung a day earlier. Aside from Korea and Japan, the United States and Canada are also participating in the GTAC program.
The inclusion came amid growing concerns over powerful AI models, such as OpenAI’s GPT-5.5 and Anthropic’s Mythos, being misused to pose cybersecurity threats to critical social infrastructure. Acknowledging the risks associated with their advanced models, OpenAI and Anthropic are operating the Daybreak initiative and Project Glasswing, respectively, providing early access to their latest models to a limited number of trusted partners so they can strengthen their cyber defenses.
As Korea gains access to OpenAI’s Daybreak initiative, the possibility of the country also joining Anthropic’s Project Glasswing is increasingly coming into view.
Source: Korea Times News
